...
If Navi really was about chasing the low-end of the market, I wonder why the heck they included PCIe 4
One reason is that PCIe 4 lines with their far more affordable CPUs too. Not just trying to sell GPUs, but systems.
Yes super budget folks only want to buy components for the sunk cost base system the own. Well PCIe 4 card is perfectly compatible with PCI-e v3 systems.
Second, since had to do PCIe 4 work for 7nm for CPUs ... it probably didn't add any significant costs to the put in.
Third, it is a marketing checkbox that Nvidia doesn't have. ( sell to low cost buyers are "future proofed" . when do buy that upgraded base system can keep the card and it will go faster. ).
DSC is actually part of the standard (both DisplayPort and HDMI ) . GPU vendors are suppose to implement them in a timely fashion. The implementation algorithm is laid out in the standard and the implementation space on the die isn't that huge. This isn't a high cost item. If get first version 1 done before competitor can move to version 2 ( or 1.1 ) later when competitor has just got to version 1.
when neither of those are going to be relevant for a market buying budget SSDs, 1080p monitors,
budget isn't particular 1080p anymore. 1440p monitors now are were 1080p monitors were 3-4 years ago. 4K is going to be a marketing checkbox in the current context.
Exactly why I think this is a low point for them. I’m not saying it can’t or won’t get better, but right now it certainly feels like AMD would have to actively sabotage themselves to do worse in the GPU space.
A contributing factor here is the hooks from Navi into the PS/5 and Xbox Series X. If Xbox Series X isn't a flop then this is more so slow out of the gate. If both do reasonable well then being stretched thin in late 2019 through now will pay off. If both don't do well then very substantive parts of Navi would have turned out to be a distraction.
The other contributing factor is having to weave the macOS drivers updates into the macOS release process. 10.15 was a crap storm on the initial iterations. It can't imagine that actually helped AMD's development process for the second half of 2019. Apple can be kind of cheap so I highly doubt they compensated AMD for crapping on the release process either. Drivers for the 5300M and 5500M probably had priority ( Mobile Macbook Pro over Mac Pro or eGPUs in priority. )
It would be kind of curious as to where AMD has put their driver developers too. ( if somewhere in a Covid-19 crap storm this Winter that wouldn't help either. There is a decent chunk of graphics development done in Shanghai, among other places. )
I like AMD and what they are trying to do, in general. But they come across as a company that has far too much experience tripping over their own feet, and are doing an awful lot of it in the last year or so.
AMD certainly has bugs to fix. But Apple has gobs of them too. Putting them together on a shared project probably isn't better.
I think AMD is doing better than there were 4-5 years ago, but also covering more areas so more places to have bugs.
Unless bring the depth and breadth of the stack being developed into perspective than can get mired in the simplistic metrics as number of bugs.
www.amd.com
